Authors:
Manabu Kitagata
and
Jun-ichi Inoue
Affiliation:
Hokkaido University, Japan
Keyword(s):
Genetic algorithms, Evolutionary optimization, Machine learning, Population dynamics, Thermodynamics, Average-case performance, Spin glass model, Statistical physics.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biocomputing and Complex Adaptive Systems
;
Co-Evolution and Collective Behavior
;
Computational Intelligence
;
Evolutionary Computing
;
Evolutionary Multiobjective Optimization
;
Genetic Algorithms
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Machine Learning
;
Soft Computing
;
Symbolic Systems
Abstract:
A general procedure of average-case performance evaluation for population dynamics such as genetic algorithms (GAs) is proposed and its validity is numerically examined. We introduce a learning algorithm of Gibbs distributions from training sets which are gene configurations (strings) generated by GA in order to figure out the statistical properties of GA from the view point of thermodynamics. The learning algorithm is constructed by means of minimization of the Kullback-Leibler information between a parametric Gibbs distribution and the empirical distribution of gene configurations. The formulation is applied to a solvable probabilistic model having multi-valley energy landscapes, namely, the spin glass chain. By using computer simulations, we discuss the asymptotic behaviour of the effective temperature scheduling and the residual energy induced by the GA dynamics.